Data, Demand, and Demographics: A Symposium on Housing Finance

Co-presented by the Urban Institute and CoreLogic

November 2, 2016

Welcome  Laurie Goodman – Director, Housing Finance Policy Center, Urban Institute  Faith Schwartz – Principal, Housing Finance Strategies, representing CoreLogic

Join The Conversation #HousingData 2

Urban Challenges and Opportunities Policies of Permitting, Preserving, and Advancing our Cities



Ed Glaeser – Fred and Eleanor Glimp Professor of Economics, Harvard University

Join The Conversation #HousingData 3

New (and Old) Directions in Housing Market Research Edward L. Glaeser Harvard University

4

Is the Bubble Back? (S+P, C-S, Corelogic)

5

Three Themes • The Rise of Sub-City Data and Neighborhood Measurement Tools (with Cesar Hidalgo, Nikhil Naik, Scott Kominers) • The Growing Academic Consensus on Real Estate Bubbles (with Charles Nathanson) • The Underappreciated Power of Supply (with Joe Gyourko, also Chinese material joint with Andrei Shleifer, Yueran Ma, Wei Huang)

6

FHFA Zip Code Data From Beogen, Derner And Larson

7

Hipsman (2015) Using Zillow Sub-city

8

The Promise of Google Street View + Computer Vision (with C. Hidalgo, N. Naik and S. Kominers)

• Google Street View has covered more than 3,000 cities from 100 countries across the world • High resolution imagery at street-level: amenable for analysis by both humans and computers • Time series: 2008 - Present • Data from India and China should be available soon (from various providers), already available for Brazil, Indonesia etc.

9

The Promise of Google Street View + Computer Vision

• Availability of Street View paralleled by impressive gains in computer vision technology, fuelled by deep learning. • Opportunity to develop automated surveys of the built environment at unprecedented resolution and scale

10

Streetscore (Cesar Hidalgo): How safe does this place look to humans?

1.8/10

9.2/10

Goal: Train a computer to assign a score to a street view image for “perceived safety” from image pixels Can be extended to “perceived” wealth, liveliness, cleanliness etc.

11

Train A Computer Vision Model to Predict Streetscore (Perceived Safety FROM Naik) Training Examples

8/10

Computer Vision

Predicted Streetscore

5/10

3/10

Image Features Derived from Pixels

6.4/10

12

Computing Urban Change

• Urban Change Coefficient (UCC) : Change in Streetscore of images of the same location between 2007 and 2014

1.8/10

9.2/10

UCC = +7.4 (positive change) Naik et.al., NBER Working Paper, 2015

13

Urban Change Coefficient Significant decay

14

Urban Growth in New York City 2007 - 2014

15

Which demographic factors precede physical urban change? With Hidalgo, Kominers, Naik and Raskar (2015)

• 5 cities – 2,514 census tracts • Socioeconomic data from Census • Multivariate spatial regressions

16

17

18

High Frequency Momentum, Low Frequency Mean Reversion

19

Beliefs follow Past Price Growth

20

Irrationality Just Fits the Data Better

21

Figure 13: Mean Reversion across Zip Codes (Mean Residuals from Hedonic Regression) .

Log Value Change 1928-1935 0

10040

10012

10010

-.2 10009

10039 10023 10032 10011 10026 10017 10003 10030 10036 10021 10029 10025 10016 10027 10024 1000210035 10128 10028 10033 10019 10007 10037 10075 10031 10013 10022 10014 10001 10065

10006

10018

10034

-.4

10004

10005

10038

10020

-.6 .1

.2

.3 Log Value Change 1921-1928

.4

.5

Source: Data from Nicholas and Scherbina (2011)

22

.25

Change in Housing Prices, 2001-2006 vs. 2006-2011

Change in FHFA Price, 2006-2011 -.75 -.5 -.25 0

Houston New York DC Detroit Phoenix

-1

Las Vegas

0

.2

.4 .6 Change in FHFA Price, 2001-2006

.8

23

Multi Family Permits Single Family Permits

24

Nov 2014

Sep 2013

Jul 2012

May 2011

Mar 2010

Jan 2009

Nov 2007

Sep 2006

Jul 2005

May 2004

Mar 2003

Jan 2002

Nov 2000

Sep 1999

Jul 1998

May 1997

Mar 1996

Jan 1995

Nov 1993

Sep 1992

Jul 1991

May 1990

Mar 1989

Jan 1988

Nov 1986

Sep 1985

Jul 1984

May 1983

Mar 1982

Jan 1981

Nov 1979

Sep 1978

Jul 1977

May 1976

Mar 1975

Jan 1974

Nov 1972

Sep 1971

Jul 1970

May 1969

Mar 1968

Jan 1967

Nov 1965

Sep 1964

Jul 1963

May 1962

Mar 1961

Jan 1960

Single Family and Multi-Family Permits Over Time

2,000

1,800

1,600

1,400

1,200

1,000

800

600

400

200

0

25

26

27

28

29

30

31

The Great Chinese Housing Boom

People lining up outside a residential project before sales start. Hefei, Anhui Province.

32

33

34

35

36

37

Data, Demand, and Demographics: A Symposium on Housing Finance

Co-presented by the Urban Institute and CoreLogic

November 2, 2016

Housing and Economic Outlook 2016 and Beyond

 Frank Nothaft – Senior Vice President and Chief Economist, CoreLogic

Join The Conversation #HousingData 39

Housing and Economic Update: 2016 and Beyond Data, Demand, and Demographics: A Symposium on Housing Finance Frank Nothaft, CoreLogic SVP & Chief Economist November 2, 2016

40 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

The ‘New Normal’ 1. Low mortgage rates are the norm 2. Household composition increasingly diverse 3. Sales rise but turnover remains below ‘average’ 4. Originations: Purchase & HELOC up, Refi down 5. Loan performance excellent (new credit ‘lower risk’) 41 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

When Will the Fed Raise Target Rates? (Soon) Federal Funds Target (interest rate, in percent) 4.0 3.5

Minutes Sep 21, 2016 Sep 21, 2016 Median

3.0 2.5 2.0 1.5 1.0 0.5 0.0 Sept. 21, 2016 Median

2016

2017

2018

0.625

1.125

1.875

2019 2.625

Longer Run 2.875

Source: Federal Open Market Committee Meeting on September 21, 2016. In the plot each circle indicates the value (rounded to 42 the nearest 1/8 percentage point) of an individual FOMC participant’s judgment of the appropriate level of the target federal funds rate at the end of the specified calendar year or over the longer run. ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Low Mortgage Rates Are the New Norm Interest Rate on 30-Year Fixed-Rate Mortgages (percent) 7%

Forecast

6%

Dec. 2017:

5%

4%

4.2%

Great Recession

3% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: Freddie Mac Primary Mortgage Market Survey®, IHS Global Insight October 2016 projection.

43 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Large Demographic Tailwind Has Arrived Population in 2015 (Millions)

Largest Age Cohort

4.8 Average Age Firsttime Homebuyer

4.6 4.4

Average Age Repeat Buyer

4.2 4 3.8 3.6 3.4

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

Age in 2015 Source: U.S. Census Bureau, Population as of July 1, 2015

44

©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Household Composition More Diverse Three-fourths of New Households Will Be Minority-Headed 2015 Share of All Households

2015-25 Share of Household Growth Asian & Other 18%

Hispanic Asian & Other 7% 13% Black 12%

White 68%

117 Million Households in 2015

Hispanic 40%

White 24%

Black 18%

12 Million Increase by 2025

Source: Census Bureau Housing Vacancy Survey (2015 household count), Harvard University Joint Center for Housing Studies (Baseline Household Projections for the Next Decade and Beyond, Working Paper w14-1)

45

©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Sales Rise but Home “Turnover” Remains Low Does ‘New Normal’ Have Lower Turnover? Home Sales as a Percent of Housing Stock 8%

6%

2000-2003 average = 5.6% 4%

2%

0% 2000

2002

2004

2006

2008

2010

2012

2014

2016

2018

Source: CoreLogic REAS MarketTrends through June 2016, Census Bureau HVS, Forecast averages 46 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential. projections of FNMA, FHLMC, Zelman and Associates, MBA, NAR and NAHB.

46

Americans Are Keeping Their Homes Longer Number of Years A Home Is Owned (Median) 14 12

Owner Occupants

10

Home Sellers

8 6 4 2 1985

1990

1995

2000

2005

2010

2015

Source: American Housing Survey for the United States, various years (difference between survey year and median year owner-occupant moved into unit), CoreLogic public records for 47 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential. United States (length of time between recorded sales on same home).

Low For-Sale Inventory: Part of a ‘New Normal’? Homes-For-Sale Inventory as a Percent of Households 4%

3%

2%

1%

0% 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Sources: National Association of Realtors, U.S. Census Bureau (New Residential Sales and Housing Vacancy Survey). Note: Existing home inventory excludes Condo & Co-op Inventory before 1999.

48

©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

U.S. Home Prices: 5% Below 2006 Peak Projected to Return to Peak by Early 2018 CoreLogic Home Price Index (January 2000 = 100) 220

-- Forecast --

200

5% 180 160

43%

140 120 100 2000

2002

2004

2006

2008

Source: CoreLogic Home Price Index (November 1, 2016 release)

2010

2012

2014

2016

2018

49 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Price Growth Faster For Lower-Priced Houses Cumulative Price Growth Through July 2016 (percent) Since July 2015 8%

Since March 2011 60%

6%

45%

4%

30%

2%

15%

0%

0% More Than 25% Below Median

25% or Less Below Median

Price Growth Since:

Up to 25% Above More Than 25% Median Above Median

One Year Ago

Source: CoreLogic HPI, Single-family Detached (November 1, 2016 release); March 2011 is “Post-Great Recession” price trough.

Price Trough

50 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Originations: More Purchase, Less Refi in 2017 Single-family Mortgage Originations (Billions of dollars) $2,400 Est. --Forecast--

$2,000 $1,600

Refinance

$1,200 $800

Purchase

$400 $0 2009

2010

2011

2012

2013

2014

2015

2016

2017

Source: Originations are an average of the latest projections released by Mortgage Bankers Association, Fannie Mae, Freddie Mac and Zelman & Associates. Fannie Mae as of October 2016. Zelman, Freddie Mac and MBA forecast as of September 2016. 51 2009-2014 are benchmarked to HMDA. Numbers do not include HELOCs. ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

HELOC Volume Up in 2016 Approved HELOCs (Billions of Dollars) $400 $350 $300 $250 $200 $150 $100 $50 $0 2000

2004

2008

2012

2016 (Through August, Annualized)

Source: CoreLogic public records, second-lien HELOCs placed more than 60 days after first lien.

52

©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Mortgage Credit Risk Along Six Dimensions First-Lien Purchase Money Originations 300

Credit Score Less Than 640

200 Low & No Doc Share

LTV Share 95 And Above

100

0

DTI Share 43 And Above Condo Co-op Share Source: CoreLogic Loan Servicing Database

Benchmark (2001 and 2002 Originations) Non-Owner Occupancy Share

Current (2016:Q2) 53

Source: CoreLogic

©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Excellent Loan Performance: Part of a ‘New Normal’? Serious Delinquency Rate by Origination Cohort 1999-2003

2004-2008

2009-2014

Source: CoreLogic: March 2016 54 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

The ‘New Normal’ 1. Low mortgage rates: Below 5% next two years 2. Household composition increasingly diverse 3. Sales rise but turnover remains below ‘average’ 4. Originations: Purchase & HELOC up, Refi down 5. Loan performance excellent (new credit ‘lower risk’) 55 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Where to find more information Look for regular updates to our housing forecast, commentary and data at http://www.corelogic.com/blog @CoreLogicEcon @DrFrankNothaft

The views, opinions, forecasts and estimates herein are those of the CoreLogic Office of the Chief Economist, are subject to change without notice and do not necessarily reflect the position of CoreLogic or its management. The Office of the Chief Economist makes every effort to provide accurate and reliable information, however, it does not guarantee accuracy, completeness, timeliness or suitability for any particular purpose. CORELOGIC and the CoreLogic logo are trademarks of CoreLogic, Inc. and/or its subsidiaries. 56 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.

Data, Demand, and Demographics: A Symposium on Housing Finance

Co-presented by the Urban Institute and CoreLogic

November 2, 2016

Panel One

Integrated Services and Inclusionary Housing for Changing Demographics  Ellen Seidman – Senior Fellow, Urban Institute (Moderator)  Nela Richards – Chief Economist, Redfin  Rolf Pendall – Director, Metropolitan Housing & Communities, Urban Institute  Jim Carr – Visiting Fellow, Roosevelt Institute  Susanne Slater – President and CEO, Habitat for Humanity of Washington, D.C.  Judi Kende* – Vice President, Enterprise Community Partners

Join The Conversation #HousingData 58

Integrated Services and Inclusionary Housing for Changing Demographics: Can we build our way out of this? Urban Institute/Core Logic Demand, Data and Demographics Symposium November 2, 2016 Nela Richardson, PhD Chief Economist, Redfin Corp.

59

Five key research findings 1. The housing market is chronically undersupplied (and the rent is too darn high!) 2. Economic mobility depends on geography 3. Economically integrated neighborhoods are rare 4. Land-use regulation affects economic inequality 5. Zoning threatens U.S. productivity and economic growth

60

61

Integrated cities are rare City

Balanced Mix Area

High-end Area

Affordable Area

Boston

51%

35%

15%

Seattle

31%

10%

59%

Washington, DC

30%

25%

45%

San Jose

24%

53%

24%

Denver

24%

7%

69%

San Diego

20%

40%

40%

Los Angeles

19%

74%

7%

Chicago

17%

5%

79%

Austin

16%

11%

73%

Phoenix

13%

11%

76%

Houston

12%

16%

72%

Philadelphia

11%

6%

82%

Baltimore

11%

3%

86%

San Francisco

10%

88%

2%

San Antonio

8%

5%

88%

Memphis

8%

4%

88%

Jacksonville

7%

3%

90%

Detroit

7%

1%

92%

Indianapolis

6%

2%

92%

Columbus

4%

1%

95%

62

Job accessibility in Chicago

63

Families want access to highly ranked schools

64

…and walkable communities

65

Walkable communities are highly valued

66

67

Families are moving farther from the city center

68

69

Thank you! Redfin Research https://www.redfin.com/blog [email protected] @NelaRichardson

70

Demographics, homeownership, and home equity Trends and policies for high- and low-cost states Rolf Pendall, Ph.D. Co-Director, Metropolitan Housing & Communities Policy Center November 2, 2016

71

72

Homeownership falling, faster in high-cost states Per capita homeownership rates, observed and projected to 2040 75%

50%

25%

0% CA

TX

White NH

CA

TX

Black NH 2010

2020

CA

TX

Other NH 2030

CA

TX

Hispanic

2040

Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. 72

73

Home equity falling in CA, rising in TX Median home equity per capita, homeowners, projected to 2040 (2015 $000) $150 $125 $100 $75 $50 $25 $0 CA

TX

White NH

CA

TX

Black NH 2010

2020

CA

TX

Other NH 2030

CA

TX

Hispanic

2040

Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. Dollars are wage-adjusted 2015 values. 73

74

Seniors’ home equity threatened in high-cost states Median per capita home equity (2015 $000) at age 75

Homeownership at age 75

$250

100% 75%

TX CA

$200 $150

50%

CA

$100 25%

TX

$50 $0

0% 2010

2020

2030

2040

2010

2020

2030

2040

Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. Dollars are wage-adjusted 2015 values. 74

75

Policy implications: Federal, state, and local No silver bullets: we need every solution For people, we need policies that will  raise incomes and wages over the life course in all states  reduce income insecurity in all states

For housing, we need federal, state, and local actions to  Guarantee stable affordable housing in safe neighborhoods for extremely low income people in all states  For high-cost states especially:  Boost housing supply by reducing regulatory burdens and investing in infrastructure  Facilitate access to ownership for younger households  Facilitate efficient use of homes and lots by seniors  Phase out/redirect mortgage interest and property tax deductions 75

Homeownership and Household Wealth

James H. Carr Coleman A. Young Chair and Professor In Urban Affairs Wayne State University And Visiting Fellow, The Roosevelt Institute

At the Urban Institute

Annual Urban/CoreLogic Symposium Washington, DC November 2, 2016

76

1. Diverse populations 2. Diverse land use within same community 3. Walkable communities 4. Access to mass transit 5. Diverse housing stock 6. Boutique restaurants and retail/rich nightlife 7. Historic landmarks and art and cultural institutions 8. Universities and other centers of learning 9. Centralized location within metropolitan areas 10. Often bordered by impressive waterfronts

Source: Current Population Survey/Housing Vacancy Survey, 2001–14; Survey of Consumer Finances, 2001– 13); Wall Street Journal.

77

1. Diverse populations 2. Diverse land use within same community 3. Walkable communities 4. Access to mass transit 5. Diverse housing stock 6. Boutique restaurants and retail/rich nightlifeOften bordered by impressive waterfronts

Source: Author’s calculations of HMDA data, 2000–14.

78

Source: Current Population Survey/Housing Vacancy Survey, Source: eMBs, CoreLogic, HMDA, IMF, Urban Institute.

79

Source: Urban Institute calculations from HM DA and CoreLogic data. Note: Shares are computed within each race and ethnicity group. Declines are the percent decline in loans from 2001 to 2013.

80

Source: www.supercomputinginengineering.com

81

.

Source: CFED, Institute for Policy Studies; Wall Street Journal.

82

Source: The Racial Wealth Gap: Why Policy Matters. Demos and the Institute on Assets and Social Policy. Washington, DC. 2015.

83

Source: The Racial Wealth Gap: Why Policy Matters. Demos and the Institute on Assets and Social Policy. Washington, DC. 2015.

84

•

Require all federal mortgage agencies to use the most current and predictive credit scoring models on the market

•

Eliminate GSE loan level (risk-based) pricing

•

Return GSE g-fees and FHA MMPs to levels that reflect future projected losses

•

•

•

•

Leverage distressed property sales by all federal housing agencies to better promote affordable homeownership opportunities Hold private lenders accountable for exclusionary lending practices Allow the GSEs to reserve for future losses or establish a Treasury fund (within the conservatorship framework) for losses GSE losses Reform the housing finance system to address the multifaceted housing and community investment needs of America’s distressed communities into the 21st Century 85

86

Susanne V. Slater, President & CEO Habitat for Humanity of Washington, D.C.

WHY HOUSING EQUITY MATTERS 87

www.dchabitat.org

87

What Is DC’s problem? • As in other U.S. cities, market forces are driving inmigration of younger, white, and more highly-educated populations and forcing out-migration of low- to moderate-income people of color • Very limited land boundaries accelerate problem because moving further out means moving out of the city altogether

Source: “DISTRICT OF CHANGE: GENTRIFICATION AND DEMOGRAPHIC TRENDS IN WASHINGTON, DC” Chicago Policy Review, July 23, 2014 88

www.dchabitat.org

88

The Funding Gap

Current Housing Production Trust Fund Value as a Percent of Total Need

$100 Million

$5 Billion

Source: "Will D.C.’s Housing Ever Be Affordable Again?" The Atlantic, August 19, 2016 via D.C. Fiscal Policy Institute

89

www.dchabitat.org

89

Goal: Preserve Diversity & Inclusion

1. Maximize cost-effectiveness of solutions by anchoring affordable housing in neighborhoods prior to gentrification 2. Substantially increase low-income homeownership to address wealth gap 3. Allow low-income homeowners to reap expected gains 4. Balance homeownership with permanently-affordable rentals 90

www.dchabitat.org

90

What does Habitat do?

• 15th largest homebuilder in the U.S. • Provides affordable homeownership opportunities to low- to moderateincome families • Utilizes private, government, and philanthropic funding • Builds in pre-gentrified neighborhoods • Utilizes innovative, cost-effective approaches including Inclusionary Zoning projects, green building, and voluntourism 91

www.dchabitat.org

91

Case Study: Ivy City

• Worked with a coalition of nonprofits (DC Habitat, Manna Inc., Mi Casa), to build in the most-distressed census tract in the city • In a neighborhood with a population of